concept
active
concept:other-referencing-activationsOther-Referencing Activations
Latent model activations when processing inputs framed from another agent's perspective
Neighborhood — ranked by edge-count
Methods (1)
method
- SOO Loss FunctionaboutA loss function measuring the dissimilarity of latent model representations of self and other, minimized during fine-tuning
Concepts (1)
concept
- Self-Referencing Activationsassociated_withLatent model activations when processing inputs framed from the model's own perspective
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Internal representations of the model on which probes operate; the method uses activations to rank datapoints.
- Model-independent feature comparison based on correlating activation vectors across a fixed diverse dataset
- Intervention method that adds a learned direction vector to residual stream activations to steer model behavior
- Pearson correlation of feature activations across 40M tokens used to measure feature similarity and universality across models
- Clamping activations along the Assistant Axis to remain above a minimum threshold (25th percentile), introduced as a stabilization method
- Adding steering vector in forward direction to push model activations toward stronger reflective behavior.
- Explicit textual or graphical links between parts of a work, dynamic and virtual.
- Framework training LLMs to answer questions about externally-provided activation vectors